Alibaba Eyes Massive AI Model with 5–10 Trillion Parameters, Unveils New Chip
Alibaba is charting an ambitious course in artificial intelligence with plans to develop a foundational AI model containing between 5 trillion and 10 trillion parameters. The announcement, which positions the company among firms pursuing increasingly large-scale AI systems, was accompanied by the unveiling of a new custom chip designed to support such compute-intensive workloads.
The parameter count being reported places Alibaba's proposed model among the largest contemplated by any company to date, reflecting the ongoing trend toward scaling neural networks as a path toward improved capabilities. This development occurs as multiple technology firms race to advance their AI offerings, with model size often cited as a factor in performance.
The simultaneous announcement of a dedicated chip underscores the hardware challenges inherent in training and deploying such massive models. Custom silicon has become a strategic priority for major AI developers seeking to reduce dependence on external suppliers and optimize performance for their specific software stacks.
Details regarding the timeline for the model's availability and the chip's specifications have not yet been fully disclosed.